Firefox 3 caret problems

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Firefox 3 caret problems

Aaron Leventhal-3
Firefox 2 worked pretty much perfectly with the caret and Window-Eyes, JAWS and
ZoomText, but unfortunately that all changed with Firefox 3, because of the new
rendering architecture.

We have new IAccessible2/IAccessibleText interfaces that can resolve it once and
for all but that does not address older screen readers that currently work with
Firefox.

Today I checked in a patch to fix most of the Firefox 3 problems (the 6/15 build
will have the fix), by having an invisible system caret follow the Gecko caret
around.

Here's the current status of each AT:

Window-Eyes:
Firefox 2, caret worked completely
Before patch, caret totally broken
After patch, two problems:
1) caret sometimes beeps for a character after a single pixel "i" or "l"
2) caret sometimes repeats last letter on line, make it appear to occur twice
(pqr seems like pqrr)

ZoomText:
Firefox 2, caret worked completely
Before patch, caret totally broken
After patch, visual caret tracking completely fixed, but audio caret echo
sometimes says "blank" when user presses right arrow at end of line and then
arrows backwards

JAWS:
Firefox 2, caret worked completely
Before patch, caret sometimes (not often) says "blank" or the wrong thing on a
given character. Eventually the caret corrects itself after several left/right
key presses.
After patch, same error -- caret sometimes says "blank" or the wrong thing on a
given character. Eventually the caret corrects itself after several left/right
key presses.

The remaining problems are small, but important. We'll have to address the last
problem with each AT vendor separately. Feel free to help us test and provide
succinct, clear information to the vendors.

Ideally the vendors wil start to use IAccessible2, which is much less likely to
break than trying to map the visual caret with their offscreen model. In fact,
the offscreen model should no longer be necessary for Firefox 3 support.

- Aaron
_______________________________________________
dev-accessibility mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-accessibility
Reply | Threaded
Open this post in threaded view
|

Re: Firefox 3 caret problems

Tim Keenan
I thought the caret might work properly in NVDA, because I know Mick has
implemented some parts of IA2, but I didn't have any luck with the 05-25
snapshot.  I believe some changes he's making now will do the trick, but
haven't tried the latest.  I'll give it a try later today.


Aaron Leventhal wrote:

> Firefox 2 worked pretty much perfectly with the caret and Window-Eyes,
> JAWS and ZoomText, but unfortunately that all changed with Firefox 3,
> because of the new rendering architecture.
>
> We have new IAccessible2/IAccessibleText interfaces that can resolve it
> once and for all but that does not address older screen readers that
> currently work with Firefox.
>
> Today I checked in a patch to fix most of the Firefox 3 problems (the
> 6/15 build will have the fix), by having an invisible system caret
> follow the Gecko caret around.
>
> Here's the current status of each AT:
>
> Window-Eyes:
> Firefox 2, caret worked completely
> Before patch, caret totally broken
> After patch, two problems:
> 1) caret sometimes beeps for a character after a single pixel "i" or "l"
> 2) caret sometimes repeats last letter on line, make it appear to occur
> twice (pqr seems like pqrr)
>
> ZoomText:
> Firefox 2, caret worked completely
> Before patch, caret totally broken
> After patch, visual caret tracking completely fixed, but audio caret
> echo sometimes says "blank" when user presses right arrow at end of line
> and then arrows backwards
>
> JAWS:
> Firefox 2, caret worked completely
> Before patch, caret sometimes (not often) says "blank" or the wrong
> thing on a given character. Eventually the caret corrects itself after
> several left/right key presses.
> After patch, same error -- caret sometimes says "blank" or the wrong
> thing on a given character. Eventually the caret corrects itself after
> several left/right key presses.
>
> The remaining problems are small, but important. We'll have to address
> the last problem with each AT vendor separately. Feel free to help us
> test and provide succinct, clear information to the vendors.
>
> Ideally the vendors wil start to use IAccessible2, which is much less
> likely to break than trying to map the visual caret with their offscreen
> model. In fact, the offscreen model should no longer be necessary for
> Firefox 3 support.
>
> - Aaron
_______________________________________________
dev-accessibility mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-accessibility